Hacker News Reader

Comment by rafram | Hacker News Reader

Clickbait title. Should be more like "Box founder Aaron Levie says CEOs should use AI more and learn its limitations."

He's essentially saying that C-suite people overestimate how effective LLMs are at one-shotting hard problems, and underestimate the human maintenance work that follows.

View on HN

The quote literally from Levie is "CEOs are uniquely prone to AI psychosis because they’re sufficiently distant from the last mile of work that still has to happen to generate most value with AI".

The author calling it "psychosis" discredits the author.

Whether or not it's actually a kind of psychosis, it has become common terminology. You see it used about people that have LLM boyfriends or girlfriends, and some that have started new religions based around their LLM gods. There are also a group that believe they have "awakened" their LLMs and have instructions for others to do the same thing.

Anybody mad about dubiously co-opting jargon like “AI Psychosis” to make cool-sounding but shaky analogies probably should look into the term “AI” and all the cool sounding neurobiology jargon people dubiously co-opted to describe it.

Psychosis is a medical term for a collection of symptoms that indicate a loss of contact with reality. He posits that reality is that AI cannot automate nearly as many things as people believe it can, while C-suites ignore that reality and dive head first into belief without evidence. Tell me how using that term discredits the author.

That's not psychosis, that's just being wrong. Perhaps read something: https://en.wikipedia.org/wiki/Psychosis

"Being wrong" versus "loss of contact with reality" is simply a matter of degree.

In general, "loss of contact with reality" is well understood to be an individual experience. If you think voices are talking to you, that's schizophrenia. If you think God wrote an entire book of life advice, that's Christianity. So, kinda by definition, a large group of people cannot generally experience "loss of contact with reality", they're merely mistaken.

The alternative is to accept that 90% of humans are suffering some form or another of psychosis, because there's very little that even a majority (51%) of people agree on, and that means about half the population is wrong about any given topic. Multiply by the number of topics out there...

You've assumed an uncharitable boundary and then said "see, that's obviously wrong". Well of course it is if the boundary you selected places 90% of humanity on the side of "loss of contact with reality".

So just, you know, don't do that. Select a boundary that places 90% of humanity on the side of "wrong about a number of things but within the bounds of normalcy for the species at large", sprinkle in some caveats about "in the industrialized world" and "for someone with at least a highschool education" and such, and you've got what I was actually describing.

My point is that AI psychosis is well within the bounds you're mentioning: yeah, those people are wrong about a number of things, but it's well within the bounds of normalcy for the species at large; especially in the industrialized world; especially for anyone that has at least a high school education.

You can't cherry-pick a definition where "AI psychosis" is meaningfully different from "wrong about a few things" without hitting a huge swath of "within the bounds of normalcy for the species at large" - there's just too many people in the category

There is a huge difference between being incorrigibly wrong (delusion) and psychosis (much more encompassing). There is not a difference by degrees because being incorrigibly out of touch on a range of issues does not mean you are psychotic.

Would using anthropomorphize be better? Because that's what they are really doing. The problem is that the folks being described then go on step farther to think LLMs are a digital engineer and thus we don't need (most) devs anymore. This is what we describe as psychosis. If you have a better word for this phenomenon, I would love to hear it.

> a better word

"stupidity" sounds apropos

Personally, I am annoyed by AI responses that include courtesy and emotional words like "please" and "thank you". I'd prefer to just get simple factual answers.

who knows, may they are right, and 36 subagents can produce one AI baby in 1 week.

And maybe it's got 6 fingers on each hand.

Does that mean it can work 20% faster?

Not with all tasks, but it's bound to become an awesome speed typist.

I wasn't expecting a peer reviewed study, but I certainly was expecting more from that title

Reply on news.ycombinator.com